Rotating model during running

[Bas Van Hooren]

Hello,

I was experimenting with using AddBiomechanics and uploaded a running trial with a full-body markerset & GRF. Within the OpenSim Gui, the IK results look normal. When I process the results using AddBiomechanics, the model seems to 'flip' multiple times so that it runs backwards/sideways for a few miliseconds before continuing to run forwards.

First here an image that seems normal (although the pelvis geometry is missing)

OpenSim GUI

Normal run.JPG (17.66 KiB) Viewed 676 times

Here a few miliseconds later where the model is flipped (partly) (correct image appears below the post, at is does not seem to work placing it in line)

OpenSim GUI

Normal run.JPG (17.66 KiB) Viewed 676 times

In the GUI everything looks normal, including the GRF (although they have been filtered):

OpenSim GUI

Normal run.JPG (17.66 KiB) Viewed 676 times

Any idea on what could be causing this?

[Nicholas Bianco]

Hi Bas,

Are you running the inverse dynamics step with your data? If so, the rotating model is most likely because the dynamics fitting step is trying to adjust the kinematics to minimize large pelvis angular residuals in your data.

How large are the residual forces if you just run inverse dynamics in the OpenSim GUI?

-Nick

[Bas Van Hooren]

Dear Nicholas,

Thank you for your response and sorry for my delayed response. I have been trying to re-process the running trial with filtered GRF in AddBiomechanics for the last three days but the server keeps messaging that the trial is Queued on SLURM cluster, and this is alternated with a message that the connection is lost so I don't yet have any results from this.

I did however check the residual forces at the pelvis from ID in the OpenSim GUI, see the attached image. I don't know how to interpret these forces to check if they are (too) large and could be explaining the models behavior in AddBiomechanics?

The body mass of the individual is 72.2 kg. I filtered the GRF with a 20Hz 4th order Butterworth and the screenshot is from a 15s running trial at 2.78 m/s.

[Bas Van Hooren]

By the way, if anyone has suggestions on why the trials won't process then I would also love to hear this. I already tried making a new subject but this did not solve the problem.

[Nicholas Bianco]

Hi Bas,

These residuals seem quite large to me, the vertical force is roughly 2 bodyweights. You can use the paper by Hicks et al. (2015) as a guide for reducing residual forces. They recommend that residuals should be less than 5% of the net external load applied to the model.

The processing server issues should be resolved.

-Nick

[Bas Van Hooren]

Thanks Nicholas. Thus far the processing still shows the disconnection error, but I will update on this forum if the initial problem is solved by filtering the GRF if the data is processed within the next two days or so

[Nicholas Bianco]

Hi Bas,

You might have to click "Force Reprocess" to re-queue the job (not sure if you have tried this already).

-Nick